Luke has just about run out of electronics to do with one extension lead replaced this week and on transistor to be added to the power cut-off circuit.
Harley plans to re-work the power relay box to clean things up a bit and to ensure that all the components are rated for at least 30 amps (except for the fuses which should be exactly 30 amps :-).
Harley and David will work on the battery rack in the coming week.
Peter has built a new kernel for Kambara and refreshed/refined the procedure for down-loading it to user flash.
Peter has been working with Tim and Jordan to get the sensor software working with the WAM arm. Tim will synchronise the sensorSampler with the servo loop task (via message queues). The sampling loop can be aperiodic to provide more samples around the time when the position is recorded. The WAM controller will provide joint position, velocity, torque, and time.
Chris has been testing the reinforcement learning system with the Nomad indoor mobile robot using infra-red sensors and a camera. Learning is hampered by the unreliability of the sensors and tends to respond poorly when the robot becomes blind to an existing obstacle.
Sunil has the operatorInterface running without Java3D but not connecting to the server. The Browser Applet also works.
Chanop has an OpenGL Kambara display running which can be attached to the operatorInterface server. This may be a better solution that using Java3D and is faster as well.
Chanop has designed a simple controller for Kambara and is working on determining the control parameters.
No news to report on the PXC200F framegrabber.
Matt has the camera calibration systems working with two framegrabbers and is preparing for more experiments in the tank to measure the accuracy of the calibration.
Jordan is writing code to analyse data and to implement the Kalman filter for the WAM arm. Tim will provide a transformation matrix to map the arm joint angles to position of the sensor plate.
David has released a new carried board device interface. The 386 version for wani (and the WAM experiments) is tested and working, the PPC version needs further interrupt testing.
Some items to do: